KR101794310B1 - Object collision avoidance system at intersection using stereo camera - Google Patents

Abstract

The present invention relates to an object collision avoidance system at an intersection using a stereo camera. More specifically, a stereoscopic camera that monitors each road on a back road with three or more roads is combined to check the moving speed and direction of an object with respect to a three-dimensional space, and then to form a logical structure of a digital form with respect to the three-dimensional space. The speed and direction of the movement of a moving object on the road is calculated through information sharing to prevent collision by warning when the collision between the moving objects is predicted at the intersection of the road. The object collision avoidance system includes a control part, a plurality of stereo camera parts, a moving object detection part, and a memory part.

Description

{Object collision avoidance system at intersection using stereo camera}

The present invention relates to an intersection moving object collision avoidance system using a stereo camera, and more particularly, to a system and method for preventing an intersection moving object collision prevention system using a stereo camera by combining a stereo camera that views each road on a back road with three or more roads, After recognizing the direction, we form a logical structure of digital form in 3D space and calculate the speed and direction of the movement of the moving object on the road through information sharing and warn when the collision between moving objects is predicted at intersection of road And more particularly, to a system for preventing collision of an intersection moving object using a stereo camera for preventing collision.

In 2016, the percentage of pedestrians in traffic accidents is about 40%, and the proportion of accidents at intersections is about 45%.

Thus, it can be seen that accidents occur frequently between the left or right vehicle and the pedestrian or the vehicle, which are not well recognized by the human vision at the intersection.

In order to prevent this, a convex mirror may be installed at the intersection point or a speed limiter may be installed before the intersection point to slow down the vehicle speed.

However, in the case of a convex mirror, it is difficult to prevent a pedestrian and a driver of the vehicle from seeing it, and even if the visual field is blurred, it is difficult to prevent it because the field of view is not visible in the mirror.

In addition, it is difficult to prevent accidents in the case of pedestrians in the case of the speed limit bust.

In order to solve this problem, there are techniques for detecting and notifying other vehicles or pedestrians in an intersection when an overspeed vehicle is detected.

However, since most of these technologies measure the speed of the vehicles passing through the sensor and determine only the speed of the vehicle at the moment, it is difficult to continuously detect the vehicle passing through the road, It can not be judged if there are several variables.

Also, pedestrian speed has a limit that can not be judged.

Korean Patent Registration No. 10-1139733 (Apr. 18, 2012) 'System and method for preventing collision of an automobile on an intersection'

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems in the prior art, and it is an object of the present invention to provide a stereoscopic camera which combines stereoscopic cameras looking at each road on a back road with three or more roads, And direction, and form a logical structure of digital form about 3-D space, and calculate the speed and direction of the movement of the moving object on the road through information sharing, and when the collision between moving objects is predicted at the intersection of the road, The object of the present invention is to provide an anti-collision system for an intersection moving object using a stereo camera that prevents collision.

A plurality of stereo cameras 120 connected to the control unit 110 and oriented in different directions at an intersection, the control unit 110 being a main control unit (MCU) A moving object sensing unit 130 connected to the control unit 110 and sensing and classifying a moving object entering the intersection through the machine learning technique from the photographing information of the stereo camera 120, A spatial data synthesizer 140 connected to the controller 110 and determining a speed and a direction of the moving object using a depth map generated according to an imaging area of the stereo camera 120, An inter-moving-object collision prediction unit 150 for predicting the possibility of collision between moving objects from the information provided by the combining unit 140 and a moving-object inter-collision estimating unit 150 connected to the control unit 110, A crash prevention signal unit 160 for controlling the output unit 190 to output a warning sound and a warning message through the output unit 190 when there is a possibility of collision based on the collision prediction information provided by the control unit 110, A memory unit 170 for storing and updating image and processing values and an input unit 180 for a user interface connected to the control unit 110 and an output unit 190 for outputting a warning sound or warning text, 1. A collision avoidance system for an intersection moving object using a camera, comprising:

The stereo camera 120 is installed in a post (POST), and the POST is installed to be diagonally symmetrical to an edge of an intersection. On the POST, a fixed block 200 is fixed to a plurality of fixing bolts The fixing block 200 is fixed on the guide plate 220 in the shape of 'A' or 'D', and can be moved along the guide plate 220 in such a manner as to slide left and right. A rectangular through hole 230 having a predetermined size is formed on the guide plate 220 so that the fixing bolt 210 can pass therethrough and the camera fixing block 300 is inserted into the guide plate 220 to be slidable And a locking part 310 protrudes from a rear surface and a lower end of the camera fixing block 300 so as to be installed. The locking part 310 has a rectangular shape, (CA1) and a right camera A coupling groove 320 is formed in the coupling groove 320 to fix the pair of stereo cameras 120. The coupling hole 330 is screwed into the coupling groove 320, A camera fixing groove 334 is formed in the fixed fixing piece 332 and the camera mounting groove 334 is formed in the camera mounting groove 334. The camera fixing groove 334 is formed in the camera mounting groove 334, And a tightening hole 340 is further formed in the camera fixing block 300 and a tightening hole 342 is screwed to the tightening hole 340. [ The present invention provides a system for preventing collision of an intersection moving object using a stereo camera.

According to the present invention, a stereoscopic camera looking at each road on a back road having three or more roads is combined to determine the moving speed and direction of an object in a three-dimensional space, and then a logical structure of a digital form is formed In addition, it is possible to obtain the effect of avoiding collision by warning when the collision between moving objects is predicted at the intersection of the road by calculating the speed and direction of the movement of the moving object on the road through information sharing.

1 is an exemplary block diagram of a system according to the present invention.
2 is a schematic view showing an example of installation of an intersection of a stereo camera constituting a system according to the present invention.
3 is an exemplary diagram illustrating an example of an output unit constituting the system according to the present invention.
FIG. 4 is a conceptual diagram showing an operation example of a spatial data synthesizing unit constituting a system according to the present invention.
FIG. 5 is a diagram illustrating an example of collision budget calculation of a collision prediction unit between moving objects constituting the system according to the present invention.
6 is an exemplary view for explaining a fixing structure of a stereo camera constituting a system according to the present invention.
FIG. 7 is an exemplary view showing an installation example of a stereo camera in FIG.
FIG. 8 is a cross-sectional view of a main part showing an installation example of an image storage camera in FIG.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Before describing the present invention, the following specific structural or functional descriptions are merely illustrative for the purpose of describing an embodiment according to the concept of the present invention, and embodiments according to the concept of the present invention may be embodied in various forms, And should not be construed as limited to the embodiments described herein.

In addition, since the embodiments according to the concept of the present invention can make various changes and have various forms, specific embodiments are illustrated in the drawings and described in detail herein. However, it should be understood that the embodiments according to the concept of the present invention are not intended to limit the present invention to specific modes of operation, but include all modifications, equivalents and alternatives falling within the spirit and scope of the present invention.

1, an intersection moving object collision avoidance system using a stereo camera according to the present invention includes a control unit 110 which is an MCU (Main Control Unit), and a control unit 110 which is connected to the control unit 110, A moving object sensing unit 120 connected to the control unit 110 for sensing and classifying a moving object entering from the shooting information of the stereo camera 120 into an intersection through a machine learning technique, A spatial data synthesizer 140 connected to the controller 110 for determining a speed and a direction of the moving object using a depth map generated according to an imaging area of the stereo camera 120, A moving object collision predicting unit 150 connected to the control unit 110 and predicting whether there is a possibility of collision between moving objects from the information provided by the spatial data combining unit 140, Based on the collision prediction information provided by the moving object inter-collision predicting unit 150, a collision preventing signal for controlling the output unit 190 to output a warning sound and a warning message, A memory unit 170 for storing and updating image and processing values in accordance with a control signal of the controller 110 and an input unit 180 and a warning sound unit 160 connected to the controller 110. [ Or an output unit 190 for outputting a warning message.

In this case, all the remaining components except for the stereo camera 120 and the output unit 190 are mounted on the board in the form of a module to form a controller, and a POST (see FIG. 5) And can be configured to be operated by a computer.

The stereo camera 120 applies a human visual system that obtains distance information by appropriately matching two images acquired at different positions. The stereo camera 120 extracts features from the acquired image, estimates displacement through stereo matching, And the like.

An intersection moving object collision avoidance system through such a system can be implemented in the following manner.

First, the stereo camera 120 may be installed as illustrated in FIG.

6, the stereoscopic camera 120 includes a plurality of fixed bars S, and the stereoscopic camera 120 includes a plurality of fixed bars S, for example, And the stationary bar S may be fixed using a support bar or the like not shown, which is used when a signal lamp is installed.

Then, the stereo camera 120, which provides the left eye image and the right eye image, can take pictures of all directions.

Here, the illustrated example illustrates an example in which four stereo cameras 120 are installed by exemplifying an intersection of four distances, but it may be implemented in the same manner at three distances, five distances, etc., so that three or five A stereo camera 120 may be provided.

In particular, as shown in FIG. 3, a speaker SPK and a display panel (DIS) constituting the output unit 190 are provided below the stereo camera 120, preferably at the upper end of the post (POST) (SPK) or a warning message output via a display board (DIS).

Meanwhile, the moving object sensing unit 130 senses and classifies moving objects according to a machine learning technique while reading images captured by the stereo camera 120 in real time.

For example, a moving object entering an intersection is classified as a person, a bicycle, or a vehicle. This is done through a kind of machine learning.

That is, the machine learning method is one of machine learning algorithms for collecting and analyzing data by computer itself and predicting the future. First, a computer is learned on an algorithm basis, and new data is input to predict a result. Based on large amounts of big data to analyze the future behavior and the possibility of determining the technique.

Accordingly, if the moving object detection unit 130 learns by applying this technique, it is more effective in calculating the moving speed of the moving object, and it is possible to immediately read whether or not the moving object enters the intersection.

The spatial data composing unit 140 includes a left eye camera (left eye lens) and a right eye camera (right eye lens) in which a plurality of stereo cameras 120 are spaced apart from each other by a predetermined interval Accordingly, the number of pixels of the same point among the images photographed by the left eye lens and the right eye lens is calculated (parallax) to calculate the depth value with respect to the subject, and a depth map is generated based thereon.

In addition, when the bottom of the moving object is excluded from the detection range by using the depth map, a small object such as a leaf or a shadow moving in the vicinity of the bottom of the moving object disappears, and the moving object is mistaken as another object in the moving object detection The likelihood of detection is significantly reduced, which can significantly reduce false alarms in subsequent calculations and signals.

More specifically, as shown in FIG. 4, in the case of an algorithm for detecting an object in an image acquired using an existing monocular camera, an object is detected using a pixel value of the image, so that the shadow of the image is determined as an object There is a case.

In the present invention, since the accurate position information of the moving object can not be calculated unless the moving vehicle and the human shadow are removed, the shadow information of the object should be excluded.

In the present invention, when the position information of the moving object is calculated, the error is corrected using the matching 3D depth map through the stereo camera 120.

The 3D depth map is obtained by obtaining two images from a stereo camera in both eyes and calculating the three-dimensional position information of the image. In this case, a pair of feature points is calculated from two images and the two images are matched with each other.

Therefore, if a 3D depth map is generated, shadow information can be removed as shown in FIG. 4 without adding an additional algorithm in the calculation process.

The moving object collision predicting unit 150 matches the 3D depth map information generated by the spatial data combining unit 140 through the photographing information of the four stereo cameras 120 based on the case of the above example The speed information and the direction information of the moving object are calculated.

For example, as shown in Fig. 5, when the directions of two moving objects are confirmed and the moving speed is calculated,

ℓ _p / v _p ≒ ℓ _c / v _c = S

, It can be inferred that collision is expected after S seconds.

In this case, ℓ _p is a distance value that can be moved when the pedestrian moves at a speed of v _p , and ℓ _c is a distance value that can be moved when the vehicle moves at the speed of v _c .

This is because under the control of the control unit 110, the collision prediction unit 150 calculates the moving speed and direction of the pedestrian and the vehicle in the two-dimensional map, and when two moving objects meet at the same point at the intersection, This is expected, so it is calculated.

In this way, when the collision prediction is calculated, the control unit 110 outputs a warning sound or a warning message through the collision prevention unit 160 to prevent collision by informing the user in advance.

In addition, in the present invention, the stereo camera 120 can be installed in a posture (POST) in the same manner as in FIG. 6, so that all four directions of four distances (POST) can be monitored diagonally from the edge of the intersection .

At this time, the fixed block 200 is bolted to the post (POST) by a plurality of fixing bolts (210).

That is, the fixing bolts 210 pass through the fixing block 200 and the post (POST) at the same time, and can be installed in such a manner that a nut (not shown) is fastened to the exposed ends.

The fixed block 200 may be moved along the guide plate 220 in a manner that the fixed block 200 is fitted on the guide plate 220 and is slid to the left and right and the guide plate 220 is moved to allow the fixing bolt 210 to pass therethrough. A rectangular through-hole 230 having a predetermined size is formed.

Here, the structure in which the fixed block 200 is slidably fitted to the guide plate 220 is the same as that of the camera fixing block 300, .

In addition, the guide plate 220 may be bent in an 'a' shape or a 'a' shape so that two stereo cameras 120 may be arranged in different directions on one guide plate 220.

7, the camera fixing block 300 protrudes from the rear surface and the lower end of the camera fixing block 300 so as to be fitted in the guide plate 220, Is in the form of a crook and is engaged and fixed in a form in which the guide plate 220 is sandwiched therebetween.

In addition, a locking groove 320 is formed on the front surface of the camera fixing block 300 so as to fix the stereo camera 120, which is a pair of a left camera CA1 and a right camera CA2, The control member 330 is screwed to the fixing member 330. The control member 330 is formed with a fixed ring member 332 having a flange shape larger in diameter than the control member 330 at one end thereof, 332 are formed with a camera mounting groove 334 and one end of the stereo camera 120 of the camera mounting groove 334 is fastened.

A tightening hole 340 is further formed in the camera fixing block 300 and a tightening hole 342 is screwed to the tightening hole 340. This is because the camera fixing block 300 is disposed at a specific position The tip of the tightening tool 342 comes into contact with the one side of the guide plate 220, and the camera fixing block 300 is fixed so as not to move easily.

6, a 'b' shaped support 400 is further fixed to the bottom of the fixed block 200, and a gear installation hole 410 is vertically passed through the support 400 A rotary motor 420 is installed on the side of the gear installation hole 410 and a driving gear 430 is fixed to the motor shaft of the rotary motor 420.

A gear shaft 450 having a driven gear 440 integrally formed thereon is inserted into the gear mounting hole 410. The driven gear 440 is meshed with the driving gear 430, The lower end of the gear shaft 450 is firmly fixed to the 'C' shaped support base 460.

A shaft hole 462 is formed on both sides of the support base 460 and a shaft fixing hole 470 in the form of a bush as shown in FIG. 8 is fastened and fixed from the outside to the inside in the shaft hole 462, The speed reducer output shaft 530 is rotatably assembled to the fixture 470 and the speed reducer output shaft 530 is connected to the speed reducer 520. The speed reducer 520 is connected to the rotation shaft 510 of the tilting motor 500 do.

7 and 8, the tilting motor 500 has a structure in which the rotation shaft 510 protrudes on both sides of the rotation shaft 510. A reduction gear 520 is installed on each rotation shaft 510, And to transmit a decelerated rotational force by a reduction ratio.

In this case, when the tilting motor 500 is driven, the decelerator output shaft 530 is coupled to the shaft fixing hole 470 by spline coupling so that the reducer output shaft 530 is fixed to the shaft fixing hole 470, Is fixed to the shaft fixing hole 470, the body of the tilting motor 500 itself is rotated.

Here, the tilting motor 500 is controlled to reciprocally rotate within a range of 180 degrees, and is not rotated 360 degrees, and is controlled by the control unit 110 described above.

An image storage camera 600 is fixed to the body of the tilting motor 500.

Therefore, the image storage camera 600 can be rotated by the rotation motor 420, can be tilted up and down at the same time, and can adjust the focal distance by its own zoom in / out function.

Therefore, when the warning is performed due to a collision of the moving object, the image storage camera 600 turns the photographing direction to the collision expected point according to the control signal of the control unit 110, It can be accurately stored and used as a future reading material.

In addition, since the guide plate 220, the fixed block 200 and the camera fixing block 300, and the pedestal 400 and the support pedestal 460 must be operated externally, the durability, the water resistance, and the anti- .

For this purpose, in the present invention, they are preferably molded into the following molding composition.

The molding composition comprises 2.5 wt% of carbon black, 8 wt% of a fibrous filler coated with a metal, 2.5 wt% of aminosilane, 2.5 wt% of cyanoacrylate, 3 wt% of oxybis, 4 wt% of sodium oxide, 1.5 wt% of thiocyanium copper, 8 wt% of FT wax (Fischer-Tropsch wax), 2 wt% of sodium silicate, 2 wt% of attapulgite, 5 wt% of Pozzolan, 3 wt% of polybutene 4% by weight of monoglyceride, 2% by weight of ethylene glycol monomethyl ether, 2% by weight of sebacic acid, 2% by weight of terbium, 2% by weight of alkylene amide and the remaining polycarbonate resin.

At this time, the aminosilane is added to prevent surface cracking and inhibit the whitening phenomenon while increasing the reaction efficiency by inducing an additional reaction of the unreacted material.

In addition, the cyanoacrylate is added to enhance adhesion stability while suppressing high thermal expansion and maximizing adhesive fixing force.

The carbon black absorbs electromagnetic waves due to the dielectric properties of the carbon black as the conductive carbon black nanoparticles (powder). However, it is added in the above-mentioned range only in consideration of the moldability.

The metal-coated fibrous filler refers to Ni-CF 5%, and Ni-CF 5% refers to a fibrous filler in which 5 wt% of nickel is dispersed and coated on a carbon fiber (CF).

Such a metal-coated fibrous filler contributes to improving the durability of the resin as well as the electromagnetic wave shielding ability.

The oxybis (OXYBIS) is added to enhance the abrasion resistance by forming a cured coating through a crosslinking function, and refers to isopropyl ether (Iso-PropylEther). The sodium oxide is an alkaline oxide and is added do.

Further, the above-mentioned thiocyan copper is added as a copper-based antifouling agent in order to suppress foreign matter adhesion and to enhance moisture resistance, i.e., moisture resistance.

The F-T wax (Fischer-Tropsch wax) is added for the purpose of controlling the viscosity and increasing the crosslinkability and suppressing the deformation of the molding.

In addition, the sodium silicates are added to improve the binding between the compositions.

In addition, the atactpulse is added for thickening and bulking. However, in the present invention, it is added in order to strengthen the elasticity and increase the heat resistance according to the improvement of the elongation due to the fibrous structure.

The pozzolan is mainly used as a concrete admixture, but it is an artificial pozzolan. In the present invention, natural pozzolans collected from weathered volcanic rocks and volcanic rocks are used to increase acid resistance, corrosion resistance, durability and waterproof property, The particle size is preferably 0.1-0.2 mm.

In addition, the polybutene is added to enhance ultraviolet rays, heat and chemical stability.

In addition, the monoglyceride is added to enhance the anti-foreign material property by promoting emulsification and inhibiting foreign matter from adhering to the surface.

The ethylene glycol monomethyl ether is added to enhance the durability by maximizing the binding property between the compositions by enhancing the adhesiveness.

In addition, sebaceous acid is added to enhance the function of inhibiting foreign body adhesion.

The terbium is added as a rare earth metal belonging to the lanthanide group to enhance abrasion resistance. The alkylene amide is added in order to maintain lubricity and stability, so that the mixing is smoothly performed, and after the mixing, Lt; / RTI >

The polycarbonate resin is a base resin for enhancing abrasion resistance by establishing strong durability and hardness characteristics while ensuring transparency.

A sample of 5 cm × 5 cm × 0.5 cm size was made using the composition thus formed, and it was confirmed that the electromagnetic wave shielding ability was confirmed.

In order to confirm the water resistance (moisture resistance), the sample was kept in a state of being immersed in brine for 15 days, but there was no saline invasion.

110: control unit 120: stereo camera
130: Moving object detection unit 140: Spatial data composition unit
150: collision prediction between moving objects 160: collision prevention signal part
170:

Claims (1)

A plurality of stereo cameras 120 connected to the control unit 110 and oriented in different directions at an intersection, a control unit 110 connected to the control unit 110, A moving object sensing unit 130 for sensing and classifying a moving object entering the intersection through the machine learning technique from the photographing information of the stereo camera 120 and a moving object sensing unit 130 connected to the control unit 110, A spatial data synthesizer 140 for determining the velocity and direction of the moving object using the depth map generated by the spatial map generator 140 and the spatial data synthesizer 140 connected to the controller 110, Based on the collision prediction information provided by the moving object collision prediction unit 150 and the moving object collision prediction unit 150 connected to the control unit 110, A collision prevention signal unit 160 for controlling the output unit 190 to output a warning sound and a warning message when there is a possibility of a collision, and a storage unit 160 for storing, updating, and deleting image and process values according to a control signal of the control unit 110 A memory unit 170, an input unit 180 as a user interface connected to the control unit 110, and an output unit 190 for outputting a warning sound or a warning message, the system comprising:
The stereo camera 120 is installed in a post (POST), and the POST is installed to be diagonally symmetrical to an edge of an intersection. On the POST, a fixed block 200 is fixed to a plurality of fixing bolts The fixing block 200 is fixed on the guide plate 220 in the shape of 'A' or 'D', and can be moved along the guide plate 220 in such a manner as to slide left and right. A rectangular through hole 230 having a predetermined size is formed on the guide plate 220 so that the fixing bolt 210 can pass therethrough and the camera fixing block 300 is inserted into the guide plate 220 to be slidable And a locking part 310 protrudes from a rear surface and a lower end of the camera fixing block 300 so as to be installed. The locking part 310 has a rectangular shape, (CA1) and a right camera A coupling groove 320 is formed in the coupling groove 320 to fix the pair of stereo cameras 120. The coupling hole 330 is screwed into the coupling groove 320, A camera fixing groove 334 is formed in the fixed fixing piece 332 and the camera mounting groove 334 is formed in the camera mounting groove 334. The camera fixing groove 334 is formed in the camera mounting groove 334, And a tightening hole 340 is further formed in the camera fixing block 300 and a tightening hole 342 is screwed to the tightening hole 340. [ Moving object collision avoidance system using a stereo camera.

Images